NADIR Research
NADIR Research measures how well video-based crash reconstruction actually works, and publishes the method and the error before the product claims anything.
It publishes numbered technical reports (NADIR-TR-YYYY-NN). Each has a landing page, a PDF on this site, BibTeX, and a DOI once minted on Zenodo. Reports cite each other only where the earlier work is genuinely used. The two 2026 reports from the previous product line are kept for the record and marked archived.
Technical reports
-
A Benchmark Protocol for Measuring Error in Video-Based ADAS Crash Reconstruction
Pre-registered protocol for measuring the error of NADIR's video-based reconstruction pipeline before any accuracy figure is published: quantities, three-stage ground truth, metric definitions, reporting rules, threats to validity. Contains no results.
-
Continuous Extrinsic-Miscalibration Monitoring Above a Fleet-Visible EyeCAN Flag Baseline
Simulation study comparing NADIR's former drift-monitoring tiers against an observable EyeCAN-style binary flag baseline. Prior product line; every result comes from a simulator, not from field data.
-
Distribution-Free Fleet Monitoring of Post-Repair ADAS Extrinsic Drift Under Sparse Telemetry Proxies
Formalises post-repair drift monitoring from CAN and telematics proxies and evaluates robust-Mahalanobis, conformal, CUSUM and BOCPD detectors on a synthetic benchmark, including negative results. Prior product line; synthetic evaluation only.
Citing and reproducing
Each report page has a BibTeX entry. LaTeX sources and reproduction code are available to reviewers on request at hello@nadirai.net.